Corsair Dominator Platinum DDR3 2666 MHz 16 GB Kit

[cadaveca]

To read this review go to: http://www.techpowerup.com/reviews/C...Platinum_2666/

[erixx]

I did some AIDA64 RAM tests and my values are sames as the Samsung sticks in your reviews. I am at 1600, 1.65v and 8-8-8-8-24.

Am I wrong are is there not much to gain from upgrading mine be it this kit you reviewed or another higher speed kit...

[cadaveca]

Clockspeed can help with some workloads in a big way.

Whether those gains are worth the cost to you or not, I cannot say, but to me it is.

Those Samsung sticks are a grea tbudget buy, for sure, but at the same time, they don't ahve the LOOK that these or other sticks do. It's more than just about performance.

[Delta6326]

Great review, those sticks look amazing!

My only complaint and its not big or anything, but I'm not a fan of having speeds with 1.75v I think the max should be 1.65v. Still a great review and if these where cheaper I would buy them in a heartbeat.

[cadaveca]

Originally Posted by Delta6326

Great review, those sticks look amazing!

My only complaint and its not big or anything, but I'm not a fan of having speeds with 1.75v I think the max should be 1.65v. Still a great review and if these where cheaper I would buy them in a heartbeat.

The stock voltage IS 1.65 V.


I'm sorry if I gave the wrong impression with that.

But these are sticks meant to OC, so I checked that max during the CAS and Clock scaling testing with 1.75 V. Normal voltage for benching, IMHO, and actually, when it comes to pushing for the top scores, 1.75 V is low. Intel states 1.85 V max for hte platform, I think? I'd have to verify by checking the whitepapers.

[bpgt64]

How is price not a drawback, 750 dollars for ~5-15% gain in latency, minor gains in speed? vs a 100 dollar DDR3-1600 cas 9 kit?

[cadaveca]

Originally Posted by bpgt64

How is price not a drawback, 750 dollars for ~5-15% gain in latency, minor gains in speed? vs a 100 dollar DDR3-1600 cas 9 kit?

250 MHz over the G.Skill(just an example) kit isn't minor gains.

I see this as no different than any high-end motherboard, like the ASRock Extreme11 I reviewed a couple of weeks ago. That motherboard was $600, and the vlaue for that cost was there, and easily explained by the LSI RAID controller. Here, the added cost is covered by overclocknig headroom, esthetics, and additional functionality, like the Corsair Link connector, which allows the memory to report vital stats via software.

Hopefully I'll get a Link kit for a review soon.

I've seen the price as low as $535, and as high as $885.

That $535, I paid that for 8 GB of CAS 8 Dominator GTs.


That $100 kit isn't going to offer all that this kit does, and it's just that simple. Let me put it this way...for me, this kit allowed me to push my CPU's memory controller, and the motherboard, to the limit. I am pretty sure my CPU was what prevented me from pushing further with relaxed timings.

The only way I can confirm for sure is to get a 2800 MHz kit, and see what happens.

[mlee49]

Hey Dave, doesn't the "Platinum" heat bar across the top completely negate the airflow from the Airflow?

Seems like it wouldn't do a very good job of cooling, just running a fan into a flat wall like that.

[Widjaja]

Nice bling kit.

[cadaveca]

Originally Posted by mlee49

Hey Dave, doesn't the "Platinum" heat bar across the top completely negate the airflow from the Airflow?

Seems like it wouldn't do a very good job of cooling, just running a fan into a flat wall like that.

The Airflow fans are small, but have large, deep blades that give fairly decent pressure, thanks to the 35mm fan thickness, so yeah, it's not the most ideal, but the air definitely does flow between the sticks at least a little bit.


The LEDs in the middle of the Airflow Pro denote temperature, and the sticks remained pretty cool, as the temp LEDs stayed light blue, even with 1.75 V. The last sticks I used the display with were green at desktop with the fan, and yellow under load, so definitely much hotter.

Originally Posted by Widjaja

Nice bling kit.

Sure is. They really do not give the same impression in pictures as they do in person. Talk to Fullinfusion..he got a set after I recommended them, he's super-happy with his kit.

[EarthDog]

$750 dollars? Jesus. Will they sell these in a more monetarily appetizing 2x4Gb config?

Nice IMC you have on that CPU... bet it clocks to the moon cold (if you try!).

Originally Posted by cadaveca=

That $100 kit isn't going to offer all that this kit does, and it's just that simple. Let me put it this way...for me, this kit allowed me to push my CPU's memory controller, and the motherboard, to the limit. I am pretty sure my CPU was what prevented me from pushing further with relaxed timings.*

The only way I can confirm for sure is to get a 2800 MHz kit, and see what happens.

Id take the 2666 hyper X's over these in a heartbeat only b/c of price. The software stuff is neat.. thats it though.

but I have found that performance suffers with any motherboard when doing so. I can't honestly tell you why, but the very best performance possible, quite specifically when it comes to memory, is achieved when you stay locked to a 100.0 MHz BCLK. This has affected memory overclocking and memory purchases in a big way.

More than interested to see these details...

[cadaveca]

Originally Posted by EarthDog

More than interested to see these details...

Check the G.SKill TridentX review I posted last week, and the CAS and Clock testing there. 2500 MHz was slower than 2400 MHz.

I'm not sure why, but Copy and latency are impacted by BCLK increases on basically every board when using IVB. It's almost like some sort of divider for the controller is enabled, or something, I'm not 100% sure.


So, take your best kit of mem, one with a bit of headroom for BCLK. Bench it at XMP. bench it again @ 101 MHz. Pay attention to COPY performance first.


I've done this on every board, with many sets of memory, and each and every time, although memory speed has increased, performance in quite a few benchmarks decreases. Checking with AIDA64's memory benchmark shows drops. I'll toss up a couple of screenshots here:

Stock:


Max:



stock with max timings, plus 100 MHz CPU:

[damric]

Dave, did you happen to see what ICs are underneath?

A lot of the expensive high frequency kits are Samsung 30nm :P

[cadaveca]

Originally Posted by damric

Dave, did you happen to see what ICs are underneath?

A lot of the expensive high frequency kits are Samsung 30nm :P

Couldn't tell ya, and I think most OEMs prefer that it is not mentioned if I do discover it.

I could guess, based onteh version number, which usually indicates which ICs are used, one of the major reasons I like Corsair ram so much.

[D007]

Originally Posted by cadaveca

The stock voltage IS 1.65 V.


I'm sorry if I gave the wrong impression with that.

But these are sticks meant to OC, so I checked that max during the CAS and Clock scaling testing with 1.75 V. Normal voltage for benching, IMHO, and actually, when it comes to pushing for the top scores, 1.75 V is low. Intel states 1.85 V max for hte platform, I think? I'd have to verify by checking the whitepapers.

Yea, I use to run 2.0 v easy on my older, ddr2 ram.. 1.65-1.75 is nothing..

[yogurt_21]

not to nit pick, but why is high price not listed as a con? Otherwise good review of a nice kit.

[leeghoofd]

tWWSR value changes maybe Dave ? That's the one that cripples copy big time normally... Too bad you can't push these to the max... 2900Mhz was pretty straightforwarded with two dimms here.. 2800mhz with 4 was hard to stabilise...

I had a similar outcome with just upping the dividers... 2400 tWWSR at 5 > 2666 XMP with tWWSR at 7

Check here

[Sabishii Hito]

On X79 those sticks should do 2400 9-11-11 at 1.65v or less.

[cadaveca]

Originally Posted by leeghoofd

tWWSR value changes maybe Dave ? That's the one that cripples copy big time normally... Too bad you can't push these to the max... 2900Mhz was pretty straightforwarded with two dimms here.. 2800mhz with 4 was hard to stabilise...

I had a similar outcome with just upping the dividers... 2400 tWWSR at 5 > 2666 XMP with tWWSR at 7

Check here

Yeah, I did not play too much with tertiary timings. I know they are critical to lcocking with many DIMMs, but I do have a time budget to follow as well.

I'm not done with these sticks yet, using them for reviews now too, since they are so flexible.


I did mention that with some tweaking you can get further...it was a drop in stability that jsut could not be hcanged with primary changes or votlage, so it's definitely a timing, or my CPU that's the cause, for sure. Probably NOT the CPU. I'll have to check that out.

Originally Posted by Sabishii Hito

On X79 those sticks should do 2400 9-11-11 at 1.65v or less.

For sure!!

[leeghoofd]

X79 tests done: 2400C9-11-11 at 1.62Vdimm... 2450Mhz at 1.65Vdimm ( end of my IMC ) Gnideol even did 2400C9-11-10... impressive sticks...

https://www.dropbox.com/s/i7xcq195k3...450c9X79HP.jpg

[cadaveca]

Originally Posted by leeghoofd

2400C9-11-11 at 1.62Vdimm... 2450 at 1.65Vdimm

https://www.dropbox.com/s/i7xcq195k3...450c9X79HP.jpg

13m for 32m? WTF?




OH.

HyperPi.

[Sabishii Hito]

Originally Posted by leeghoofd

X79 tests done: 2400C9-11-11 at 1.62Vdimm... 2450Mhz at 1.65Vdimm ( end of my IMC ) Gnideol even did 2400C9-11-10... impressive sticks...

https://www.dropbox.com/s/i7xcq195k3...450c9X79HP.jpg

I have a kit of the Vengeance 2400 C9 sticks on the way, I wonder if they will be able to do 2666 C10 with 1.65v. I imagine the binning of the chips for each is about the same....

[tacosRcool]

Too expensive!

[nothappy]

Nice review dave, been looking at these badboys yesterday at my local retailer, Drooling.

The height is troublesome, I use Hyper 212 on a botom pull top push config, and my Corsair Vengeance LPs are okay(the closest dimm has a roof over it now, LOL). But with these I gotta stick with the usual horizontal Push-Pull. That will suck my airflow, and I dont have AC in my rent.

But more importantly, whats with all the miss types man? the hell is a gigabute?

[damric]

Originally Posted by cadaveca

Couldn't tell ya, and I think most OEMs prefer that it is not mentioned if I do discover it.

I could guess, based onteh version number, which usually indicates which ICs are used, one of the major reasons I like Corsair ram so much.

Of course they don't want you to know if you overpaid $700 for the same ICs found in budget RAM. Need looking heat spreaders though